The present invention relates generally to long-haul, optical fiber transmission systems. More particularly, the present invention relates to increasing the capacity of undersea cable systems.
An undersea network of fiber-optic telecommunication cables links continents, nations and lands together. Communication traffic on these submarine cables has increased dramatically in recent years and is expected to continue to grow as a result of many factors, including, for example, the globalization of world economies and businesses, the increased demand for international communications capabilities, and the development of multimedia applications and revolutionary resources such as the Internet. These factors necessitate increasing capacity not only by adding cables, but also by optimizing the operation of existing submarine cables. The term xe2x80x9ccapacityxe2x80x9d as used herein refers to the total bit rate of the cable system which is the sum of the bit rates of each channel on all the fibers within a cable.
Approaches for increasing capacity on existing cables have traditionally targeted optimizing bandwidth utilization. Long-haul optical fiber transmission paths, such as those employed by undersea cables, typically are over 1000 km in length and operate in the C-band. The C-band spans from about 1525 to about 1570 nm and is optimal for low-absorption losses. Within the C-band, capacity may be increased by increasing the number of channels and the bit rate of the channels. Most long-haul cable systems increase capacity by increasing the number channels through wavelength division multiplexing (WDM). The ultimate capacity of a WDM fiber system depends on how closely channels can be packed in the wavelength domain. Minimum channel spacing is limited, however, by the extent of interchannel crosstalk and degradation of the signal to noise ratio (SNR). Typically, channel spacing in GHz should exceed three times the bit rate in Gb/s. For example, for a bit rate of 10 Gb/s, the channel spacing should be at least 30 GHz. Therefore, for a given bandwidth, the number of channels and bit rate are interrelated and limit the capacity of a cable.
One approach to increase useable bandwidth involves using the L-band in addition to the C-band. The L-band refers to the bandwidth of about 1570 to about 1610 nm. Combining the L-band with the C-band expands the useable bandwidth from 1525-1570 (45 nm) to 1525-1610 nm (85 nm). Although using the L-band would increase bandwidth, it introduces other problems which have limited its implementation. For example, inter-channel Raman crosstalk between the C-band and the L-band tends to be significant. More specifically, in multiple channels systems, the fiber acts as a Raman amplifier such that long-wavelength channels are amplified by short wavelength channels that are within the bandwidth of the Raman gain. The shortest wavelength within the bandwidth of the Raman gain becomes the most depleted as it may pump many channels of longer wavelengths simultaneously. The signal-dependent amplification leads to power fluctuations which add to receiver noise and degrade receiver performance.
The Raman effects between the C- and L-bands are particular problematic compared to the effects just within the C-band. Given the conventional operating bandwidth within the C-band of less than 45 nm, the Raman effects between channels tend to be insignificant since the peak Raman Stokes shift is about 105 to about 120 nm from the pump wavelength. However, when the C- and L-bands are combined and form a bandwidth of close to 85 nm, the wavelength difference between the shortest and longest channels is near the peak Raman Stokes Shift.
Transmitting over the wide bandwidth of the combined C- and L-bands (85 nm) also presents problems in fiber configuration. For example, an optical fiber needs to have a low dispersion slope over the operating bandwidth such that the amount of dispersion is relatively constant across the bandwidth to minimize gain ripple. Tailoring a fiber to provide a low dispersion slope over a 45 nm bandwidth can be accomplished using known techniques such as modifying the core radius and/or index difference. However, providing a fiber having a low dispersion slope over an 85 nm wavelength (i.e., C-band+L-band) is more difficult.
Aside from dispersion, fiber loss also tends to be problematic in the L-band. Fiber loss reduces the average power reaching the receiver and results from several factors including material absorption and Rayleigh scattering. Material absorption refers to electronic and vibrational resonances associated with specific molecules within the fiber. Rayleigh scattering results from random localized variations of molecular positions in glass that create random inhomogeneities of the reflective index and act as tiny scatter centers. These fiber losses tend to be high in the L-band, which becomes particularly problematic in long-haul optical transmission paths such as undersea fibers.
Complicating high absorption losses in the L-band are the problems associated with using erbium-doped fiber amplifiers (EDFAs) in L-band amplification. One of the more significant problems with L-band EDFAs is the relatively-high pump power compared to that of C-band EDFAs which tend to have higher power conversion efficiency. Thus, not only are additional amplifiers needed to overcome the losses in the L-band, but also each amplifier requires more power than its C-band counterpart. Since power input to an undersea cable is limited due to the limited access to an undersea cable, it is extremely difficult if not practically impossible to allocate sufficient power to the required number of EDFAs for an L-band transmission over long-hauls of 6,000 to 10,000 km. Further, the additional power requirements necessitate more laser diode pumps than a comparable C-band amplifier. Aside from cost, the increased number of laser diodes also increases the size of the amplifier. It is important that the amplifiers be small enough, however, to be contained by a repeater, which, in turn, is small enough to be conveniently deployed during the cable laying process. Minimizing the size of the amplifier is also important to facilitate the use of existing manufacturing equipment and techniques (and thereby reduce costs) and to provide adequate heat dissipation by situating the heat-generating components near an outer surface.
In addition to high power requirements for L-band EDFAs, the length of the erbium-doped fiber (EDF) also tends to be significantly longer than that of the C-band counterpart. For example, a typical undersea C-band EDFA requires about a 10-20 m erbium-doped fiber (EDF), while the EDF for an L-band amplifier of comparable gain would be more than 100 m long. A longer EDF has a number of drawbacks including higher costs, temperature sensitivity, excess dispersion, polarization mode dispersion, higher noise figure, and dispersion and polarization dependent losses.
Therefore, the above problems associated with Raman effects, non-uniform fiber dispersion, fiber loss and amplification militate against the use of L-band in long-haul cable systems.
Another approach contemplated by the applicants for increasing capacity involves the use of counter-propagating signals in channels of the same wavelength, preferably within the C-band. By using counter-propagating-signals in each channel the capacity of the fiber would be effectively doubled. Despite the increase in capacity, however, applicants also recognize that the effects of backward Rayleigh scattering (BRS) over cascading amplifiers would be problematic on long-haul cable systems.
BRS is a component of Rayleigh scattering from the silica medium of an optical fiber and propagates at the same wavelength but in the opposite direction of the signal generating it. Since BRS has the same wavelength as the counter-propagating signal, eliminating or even reducing BRS is difficult and cannot be accomplished using simple and commercially-available isolators.
In cascading-amplifier configurations, the BRS begins to saturate the amplifiers and reduce the gain of the counter-propagating signals in the same channel. This results in the signal level dropping relative to BRS along the transmission path, and the SNR degrading to unacceptable levels. Therefore, applicants believe using counter-propagating signals in the same channel is not a viable approach for increasing capacity over long-haul systems.
Therefore, a need exists for increasing capacity on new and existing long-haul optical transmission paths that is cost effective and efficient while maintaining the integrity of the transmitted signal. The present invention fulfills this need among others.
The present invention provides for increased capacity on long-haul optical transmission paths, such as undersea cables, by interleaving counter-propagating signals within the C-band. The interleaving counter-propagating signals are separated by a wavelength offset which is preferably less than that required to isolate co-propagating signals using conventional wavelength filtering techniques. Rather than isolating channels based on wavelength as is done in co-propagating signals, the present invention provides for isolation of the counter-propagating signals based on direction of propagation. This way, the wavelength offset between the counter-propagating signals is less than that ordinarily used in conventional co-propagating signals. By minimizing the wavelength offset between the counter-propagating signals, the total number of channels within the C-band is increased, thus, increasing capacity within a given bandwidth.
Increasing capacity within the C-band by interleaving counter-propagating signals overcomes many of the problems described above. In particular, by keeping the bandwidth narrow, the present invention minimizes Raman effects between the signals and facilitates the use of well known and commercially-available, low-dispersion slope fibers. Additionally, by operating within the C-band, absorption losses are minimized and amplification can be performed without the need for an excessive number of C+L-band amplifiers which tend to be relatively-large (long EDF), inefficient (high power consumption) and complex (low dispersion slope over wide bandwidth) as mentioned above.
The present invention also avoids the problems associated with BRS on long-haul systems as described above by interleaving the counter-propagating signals. More specifically, the wavelengths of the BRS generated by signals moving in one direction are different from the wavelength of the counter-propagating signals. This wavelength difference allows the BRS to be filtered relatively easily compared to counter-propagating signals in the same channel where the BRS have the same wavelength and thus cannot be distinguished based on wavelength.
Therefore, by using interleaving counter-propagating signals on closely-packed channels, capacity is increased within a given bandwidth while still providing the ability to isolate the signals and reduce BRS.
Accordingly, one aspect of the present invention is a method of increasing capacity by interleaving counter-propagating signals. In the preferred embodiment, the method comprises interleaving forward-propagating signals and backward-propagating signals in forward and backward channels, respectively, within the c-band on a common optical fiber, wherein the forward-propagating and backward-propagating signals are separated by a wavelength offset that is less than that typically required for isolation between co-propagating signals. Preferably, the wavelength offset (in GHz) is about half of channel spacing of signals in the same direction. For example, if the bit rate is 10 Gb/s, and the channel spacing in same directional signals is 30 GHz, then the wavelength offset between the forward-propagating and backward-propagating signals is preferably less than 15 GHz
Another aspect of the present invention is a method for amplifying interleaving counter-propagating signals over a long-haul between cable sections.
Yet anther aspect of the invention is an amplifier system for amplifying interleaving counter-propagating signals for a long-haul cable system. In a preferred embodiment, the amplifier system comprises: (a) optics for isolating interleaved counter-propagating signals such that each signal propagates on a discrete fiber, and for combining the counter-propagating signals of each discrete fiber on a optical fiber; (b) a filter for reducing BRS on one or both of the discrete optical fibers; and (c) at least one amplifier on each discrete optical fiber for band amplifying the signals on each discrete optical fiber.
Still another aspect of the present invention is a long-haul cable system configured for interleaving counter-propagating signals. In a preferred embodiment, the system comprises: (a) a long-haul cable comprising at least one optical fiber; (b) two or more cable stations connected to the long-haul cable; and (c) a plurality of amplifier systems as described above disposed along the optical fiber.